Statistics in Nursing

Descriptive statistics are used “to synthesize and describe data” by supplying parameters commonly based on averages and percentages that are “calculated with data from a population” (Polit & Beck, 2017, p. 215). While descriptive statistics essentially describe characteristics of the data, there are levels of measurement that can be implemented to aid in that description. These levels include nominal, ordinal, interval, and ratio. Nominal level offers the least amount of detail, while interval and ratio levels offer the most. At the nominal level, variables are categorized without any obvious meaning in the relationship: for instance, age and religious affiliation would be considered nominal level variables. In descriptive statistics, nominal level data is commonly used to measure frequencies and percentages.
With ordinal level variables, there tends to be more order or meaning in the grouping, based on hierarchical arrangement: for instance, runners in a race might be ranked according to where they finished in the race. Such rankings are not used at the nominal level. At the ordinal level they are applied as labels to help establish a sense of order among the data. As with nominal level data, ordinal level is commonly used to describe frequencies and percentages.
Interval and ratio level variables provide more detail: for example, variables that have numeric values that may be added, divided, subtracted or multiplied are used at the interval and ratio levels. In descriptive statistics, the interval and ratio levels are used to describe means (averages) and/or standard deviations (Data Levels and Measurement, 2017).
Frequency distributions represent the number of occasions that a variable will hold any of its potential values within a sample. The difference between univariate and bivariate frequency distribution is that with univariate frequency distribution, data is based on a single variable with the goal being to describe that one variable: for instance, the weight of children in a specific age group. Bivariate frequency distribution occurs when data is based on two variables, the goal being to identify the relationship between the variables: for instance, the weight and gender of children in an age group. Thus, if only one variable is being determined in a data set, then univariate frequency distribution would be used. If two variables are being determined, then bivariate frequency distribution would be used.
Statistical methods are always being used in research studies. For example, Dormann et al. (2012) use descriptive analysis in their study on collinearity. They use the Pearson r-test to measure the data and they show that predictor variables of r > 0.7 “was an appropriate indicator when collinearity begins to severely distort model estimation and subsequent prediction” (p. 1). Statistical methods have thus been used to show relationships between variables, to describe how variables appear within a certain group, or to rank samples within a data set.
Key statistical tests can be used to measure significance in descriptive statistics. Tests for significance measure the probability that a relationship exists and, if a relationship exists, they measure the strength of the relationship. To test for statistical significance, a researcher must state a hypothesis, identify the null hypothesis, compute a test for statistical significance and interpret the results. These tests can include chi square test (used for nominal and ordinal level data) and t-tests (used for interval and ratio level data).
References
Data Level and Measurement. (2017). Statistics Solutions. Retrieved from
http://www.statisticssolutions.com/data-levels-and-measurement/
Dormann, C. et al. (2012). Collinearity: a review of methods to deal with it and a
simulation study evaluating their performance. Ecography, 35, 1-20.
Polit, D. F., & Beck, C. T. (2017). Generating and assessing evidence for nursing practice
(10th ed.). Philadelphia, PA: Wolters Kluwer.